The winners of the Data for Development challenge – an international research challenge using a massive anonymized dataset provided by telecommunications company Orange – were announced at the NetMob 2013 Conference in Boston last week. There were over 80 international research contributions from some of the world’s best academic institutions. The research was of an impressively high calibre and there’s a nice explanation of some of the themes that emerged on the Orange website.
In this post we’ll look at the winners and how their research could be put to use.
Best Visualization prize winner: “Exploration and Analysis of Massive Mobile Phone Data: A Layered Visual Analytics Approach”
Stef van den Elzen – Eindhoven University of Technology & SynerScope BVJorik Blaas / Danny Holten / Jan-Kees Buenen – SynerScope BVJarke J. van Wijk – Eindhoven University of Technology Robert Spousta / Anna Miao / Simone Sala / Steve Chan – Prince of Wales Fellowship at Massachusetts Institute of Technology
This internationally collaborative project developed visual analytics tools to reveal complex patterns. The winning contribution focused on tower-to-tower mobile phone traffic, operating on the assumption that call patterns change during and after major events. The team focused on changes in local call behavior and found both an increase and decrease in the number of calls over locally concentrated communication channels which strongly correlated with events. By focusing on clusters of cell towers having similar call behavior, these events can be detected. The kind of events that can be detected via changes in cell phone data include:
- weather-driven events (e.g. heavy rainfall in important cocoa areas)
- social or political events (e.g. New Year’s Eve, or large gatherings)
Enabling event detection is an important first step towards prediction, improving early intervention through development, aid and other civil initiatives.
Best Development prize winner: “AllAboard: a System for Exploring Urban Mobility and Optimizing Public Transport Using Cellphone Data”
Michele Berlingerio / Francesco Calabrese / Giusy Di Lorenzo / Rahul Nair / Fabio Pinelli / Marco Luca Sbodio, IBM Research, Dublin, Ireland
High mobile phone usage in cities present us with opportunities to track citizens’ interactions and gain insights to better plan and manage services. Transit operators can use anonymized data to map demand for travel against existing services. Currently almost all transport planning uses a four step process:
- Trip generation – How many trips are generated?
- Trip distribution – Where do the trips go?
- Mode choice – What travel mode is used for each trip?
- Trip assignment – What is the route of each trip?
This system is time-consuming, costly and the model is not sensitive to congestion effects. With large-scale data on mobility patterns, transit operators can take a data-driven and therefore more user-centric view that places the instrumented user at the center of development. In developing countries especially, performing transit analysis and optimization can have significant societal and socio-economic impact. This project – named AllAboard – is a system to optimize the planning of a public transit network using mobile phone data with the goal to improve usuage and user satisfaction.
Mobile phone location data is used to infer origin-destination flows in the city, which are then converted to ridership on the existing transit network. Travel patterns, derived from call location data, are used to propose new transit routes. An optimization model evaluates the existing transit network to increase ridership and user satisfaction, both in terms of travel and wait time. The system was tested for the city of Abidjan, Ivory Coast, to look into improving the existing SOTRA transit network. The winners demonstrated that with some minor tweaks to the bus network, costing very little money, the average commute time in Abidjan, the biggest city, could be cut by 10%.
Improving the transport infrastructure in a city like Abidjan can contribute to wider development goals – for example by making it easier for children to travel to school, helping rural communities gain access to the city to obtain employment or reducing air pollution by reducing heavy congestion.
Best Scientific prize winner: “Analyzing Social Divisions Using Cell Phone Data”
Orest Bucicovschi / Rex W. Douglass / David A. Meyer / Megha Ram / David Rideout / Dongjin Song at University of California/San Diego, La Jolla, California, USA
Looking at the volume of mobile phone activity between antennae can provide insights about the geography of social ties. This research project used the Côte d’Ivoire data provided for the D4D challenge to facilitate volume-weighted mapping of communities. In Belgium, a similar process showed that community structures are closely aligned with the French/Dutch/German parts of the country. This research project used a similar methodology to examine alignment with the 60 different local languages in Côte d’Ivoire – making things a little more complicated! The method showed that telecommunications data can provide valuable information about language communities in development settings which would be difficult to otherwise obtain. This methodology could prove very useful internationally when trying to gain information about populations to – for example – target aid or information in languages that will be understood.
First prize winner: “Exploiting Cellular Data for Disease Containment and Information Campaigns Strategies in Country-Wide Epidemics”
A. Lima / M. De Domenico / V. Pejovic / M. Musolesi, School of Computer Science, University of Birmingham, UK
A previous study by Mirco Musolesi – one of the winning team members – has shown that human movement is predictable to an extent. Of course, human mobility is one of the key factors in the spread of diseases through a population. Current disease containment strategies are usually devised on rudimentary movement scenarios. Mobility phone data provides a unique opportunity for modeling and developing strategies based on precise information about the movement of people in a region or in a country.
Using the large-scale datasets provided by Orange, taken from call patterns of a large number of individuals in Côte d’Ivoire, the team presented a model that shows how diseases spread across the country. The project simulated several epidemic scenarios and evaluated mechanisms to contain the spreading of diseases, based on the information about people mobility and social ties. An additional finding of the research was that collaborative efforts of the population can be crucial in critical scenarios: the team used the dataset to demonstrate that an information campaign based on one-to-one phone conversations among members of social groups could be an effective countermeasure. It’s easy to see how such findings could be used to improve the containment strategies of health organisations in developing countries.Image credits – main image: taken from the Best Scientific winning presentation, Visualization – from winning Visualization project, map image from Best Development prize winner presentation.
A collection of the research papers can be found here (long-read)